In the 2025 version of the Documents packages, "TimeSpan? timeout" were added to a number of interfaces, with the old versions obsoleted, for example: IWorkbookFormatProvider.Import & IWorkbookFormatProvider.Export.
This is a very strange choice, because this limits the flexibility of the interfaces for no reason at all. By only providing the TimeSpan parameter and not a CancellationToken is currently impossible to cancel the operation because e.g. an API request was canceled.
Internally these methods are implemented by first creating a cancellation token using
using CancellationTokenSource cancellationTokenSource = CancelationTokenSourceFactory.CreateTokenSource(timeout);
the token from this CancellationTokenSource is then passed to a protected method. Because this internal method uses a CancellationToken anyway, there is practically 0 development cost to exposing this in the interface, which makes the choice not to do so even more confusing.
The interfaces should expose methods that take a CancellationToken instead of a TimeSpan. This would allow for the same functionality as the TimeSpan parameter, by simply passing a cancellation token with a CancelAfter set with a TimeSpan, and an extension method could be provided for the interface which does exactly that, so users can still call these methods with a TimeSpan parameter if they wish to do so for convenience.
Please, in the next version, make these interfaces methods like this:
Workbook Import(Stream input, CancellationToken cancellationToken = default);
void Export(Workbook workbook, Stream output, CancellationToken cancellationToken = default);
and, for convenience, add extension methods for these like this:
public static class WorkbookFormatProviderExtensions
{
public static Workbook Import(this IWorkbookFormatProvider workbookFormatProvider, Stream input, TimeSpan? timeout)
{
using CancellationTokenSource cancellationTokenSource = CancelationTokenSourceFactory.CreateTokenSource(timeout);
return workbookFormatProvider.Import(input, cancellationTokenSource.Token);
}
}
Affected interfaces I've run into so far:
There may be more with this same pattern, I haven't checked.
Purpose: Long-term archiving of electronic documents with full semantic structure.
Level "1a" ensures:
Tagged PDF (with proper logical structure and reading order)
Unicode text for proper text extraction and searchability
Embedded fonts (for consistent rendering)
Restrictions:
No audio/video
No encryption
No JavaScript
No external content (everything must be self-contained)
Based on: PDF 1.4 (Acrobat 5)
PdfFormatProvider: Add support for the "Automatically resize to fit contents" table property.
The COUNTA function counts cells containing any type of information, including error values and empty text ("")
https://support.microsoft.com/en-us/office/counta-function-7dc98875-d5c1-46f1-9a82-53f3219e2509
Introduce Sanitize Document functionality.
https://experienceleague.adobe.com/en/docs/document-cloud-learn/acrobat-learning/advanced-tasks/redact
https://helpx.adobe.com/acrobat/using/removing-sensitive-content-pdfs.html
The PDF/A-1 standard uses the PDF Reference 1.4 and specifies two levels of compliance:
- PDF/A-1b - Its goal is to ensure reliable reproduction of the visual appearance of the document.
- PDF/A-1a - Its objective is to ensure that documents content can be searched and re-purposed. This compliance level has some additional requirements:
Since the PdfProcessing and its PdfFormatProvider is compliant with the PDF Reference 1.7. , the produced documents are created with this version as well: